Agate error running dbt docs generate

[friendofasquid]

Describe the bug

When running dbt docs generate, we get an error:

Tables contain columns with the same names (table_owner), but different types (<agate.data_types.number.Number object at 0x7f6112b244c0> vs <agate.data_types.text.Text object at 0x7f61128e16a0>)

Steps To Reproduce

We narrowed this down to our sources, so configure at least two source in dbt. One should have its table owner be any string (e.g. "hadoop") and the other should be an Integer (1234). Doing a CTAS with AWS Athena will, by default, give you a table with an integer for the table owner, e.g. CREATE TABLE foo AS SELECT 1 AS col1 SparkSQL CTAS will put a string in that metadata.

With those tables configured as sources, run dbt docs generate.

Expected behavior

dbt docs generate should generate the catalog.json without incident.

Screenshots and log output

If applicable, add screenshots or log output to help explain your problem.

The output of dbt --version:

installed version: 0.19.1
   latest version: 0.19.1

Up to date!

Plugins:
  - spark: 0.19.1

The operating system you're using:

Docker (python3.8-slim)

The output of python --version:

Python 3.8.9

Additional context

Add any other context about the problem here.